Adaptive thresholds is what I mean. Today I retrospectively matched AlignRT DIBH LAT- images to get the mean value of sternum displacement in the vertical direction from 25 patients at the first fractions, when DIBH DICOM reference surface was always used. The mean value was -1.5 mm, to the direction that there was a need to raise the couch to meet the (reference) sternum. At the first fraction LAT image was matched to vertebra in vertical and longitudinal directions and patients were always asked to take a BH with zero displacement in the surface vertical value in DIBH DICOM (=bar to the middle of the window). Thereby only possibility in the cases where there was a need to raise the couch to meet the sternum was instead to ask patient to take more air into lungs and to exceed the middle point of the window on the monitor as much as the error in the sternum was and acquire a new reference surface there. In several cases those systematic 1-4 mm couch corrections to meet the sternum vertical in the images were not done in the practice, since the workflow is unpractical, as mentioned. For this we could want adaptive thresholds in VRT and LNG so that patient could always take a breath to the middle of the window on the monitor.
I think most users are not doing this kind of thing. As far as I know, new reference surface for the treatment is mostly taken in the situations when users apply the small isocenter corrections, in other words shift the couch based on images (at least that was the case at the VisonRT training). Then user matches the images to the isocenter location that radiation therapist considers optimal (compromize match of the ribs and sternum), ask patient to BH into the same reference surface (deltas near zero) as at the time imaging was performed, apply the small couch shifts based on image, take a new reference surface for the treatment and let patient breath again. This works when there is need for the isocenter correction in both sternum and vertebra vertical, but if the vertebra in LAT image is in correct location already and sternum is not, we shift the vertebra out of its correct location with that kind of isocenter shift, which is not recommendable.
In our larger data this problem has been in seen. Systematic residual error for the sternum in the vertical direction has not much improved with SGRT+IGRT in comparison to SGRT only. If the sternum location is inside 4 mm in vertical in the images, we have in most cases accepted it. It is much possible that once we ask patient to exceed the window with 2 mm, she will freeze her BH to +4mm VRT and we ask her to lower the BH a little bit and she is in the middle of the window again, where she was not supposed to be if we want to correct the reference surface with +2 mm surface vertical correction. That kind of foolery is not nice.
Based on the results, the action level for sternum displacement in vertical direction has now changed into 3 mm and we try to adjust sternum such that the systematic error is as small as possible in the first place, so that the residual errors will distribute around zero instead of 1-3 mm (when there was a risk to get values 4 and 5mm in the images during the treatment course). However, for a long time we have had acceptable 5 mm realized margin needed in the sternum vertical in the image data with AlignRT DIBH, which demonstrates the accuracy in that location without IGRT.
There are several things how we can improve the thing that the sternum does not remain that shallow in the first place any more in vertical with the BH DICOM reference surface. However, during the treatment course we may face this kind of situations once in a while anyway. For those cases some kind of remedy may be desirable from the VisionRT so that there is no longer need to ask patient to do something that she does not usually do in BH, when we want to optimize the reference surface.